R

www.advmat.de

EV

Research on Advanced Materials for Li-ion Batteries

IEW

By Hong Li, Zhaoxiang Wang, Liquan Chen, and Xuejie Huang*

In order to address power and energy demands of mobile electronics and

electric cars, Li-ion technology is urgently being optimized by using alternative

materials. This article presents a review of our recent progress dedicated to

the anode and cathode materials that have the potential to fulfil the crucial

factors of cost, safety, lifetime, durability, power density, and energy density.

Nanostructured inorganic compounds have been extensively investigated.

Size effects revealed in the storage of lithium through micropores (hard

carbon spheres), alloys (Si, SnSb), and conversion reactions (Cr2O3, MnO) are

studied. The formation of nano/micro core–shell, dispersed composite, and

surface pinning structures can improve their cycling performance. Surface

coating on LiCoO2 and LiMn2O4 was found to be an effective way to enhance

their thermal and chemical stability and the mechanisms are discussed.

Theoretical simulations and experiments on LiFePO4 reveal that alkali metal

ions and nitrogen doping into the LiFePO4 lattice are possible approaches to

increase its electronic conductivity and does not block transport of lithium ion

along the 1D channel.

1. Introduction

New energy technologies are critical for the realization of anenergy future that is compatible with the goal of sustainabledevelopment. Lithium ion batteries are becoming a key-enablingtechnology for electric vehicles and hybrid electric vehicles.[1]

Since researchers at Sony Energytech developed the firstcommercial Li-ion batteries in the late 1980s,[2–4] a variety ofefforts have been undertaken to improve the battery materials.Using nanosized and nanostructured materials presents newopportunities in rechargeable Li-ion batteries for energy density,exceptionally high rate of charge and discharge, and bettercyclability.[5] Comprehensive studies of the unusual features ofthe transport and storage behaviors of ions at the nanometer scalemay lead to advanced energy storage devices.[6] In particular,spinel lithiummanganese oxide and olivine LiFePO4 are themostpromising candidates up to now for use as cathode materials ofhybrid electric (HEV) and electric vehicle (EV) batteries.[7,8] Thisarticle presents a comprehensive review of our research dedicatedto nanosized or nanostructured anode materials and themodifications of cathode materials for the next generation ofLi-ion batteries.

[*] Prof. X. Huang, Prof. H. Li, Prof. Z. Wang, Prof. L. ChenBeijing National Laboratory for Condensed Matter PhysicsInstitute of Physics, Chinese Academy of SciencesBeijing, 100190 (China)E-mail: xjhuang@aphy.iphy.ac.cn

DOI: 10.1002/adma.200901710

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinhei

2. Anode Materials

2.1. Hard Carbon Spheres (HCS)

Rechargeable lithium batteries based on theintercalation concept were first suggestedby Armand in 1972.[9] Because of safetyconsiderations, metal lithium has beenreplaced by alloys, oxides, chalcogenides,and carbonaceous materials as anodematerials.[4] In 1990, coke was used bySONY as an anode to solve the propylenecarbonate co-intercalation problem.[4] Afterintroducing ethylene carbonate into thenon-aqueous electrolyte,[10] graphite can beused properly in Li-ion batteries. Later,graphitized mesophase microbeads(MCMBs) became more popular.[11,12] Gra-phite has a theoretical capacity of 372mA hg�1 to form LiC6. Since 1990, many effortshave been made to develop high capacityanode materials to replace graphite. Non-

graphitized carbon materials, including soft carbon and hardcarbon, have been studied widely.[13–15]

Soft carbon materials show a very high reversible Li-storagecapacity but a serious voltage hysteresis during delithiation.[13–15]

Hard carbon shows a high capacity of 200–600mA h g�1 over avoltage range of 1.5–0V vs Li/Liþ. The voltage profile is mainlycomposed of two regimes, a sloped regime in a voltage range of1.0–0.1 V with a capacity around 150–250mA h g�1, and a plateauregion with a capacity around 100–400mA h g�1.[13–15] Hardcarbon materials have disadvantages such as low initial columbicefficiency and low tap density. Spherical hard carbonmaterials aredesirable, however, they are difficult to prepare through directpyrolyzing of organic or polymer precursors.

In 2000, a hydrothermal method to prepare spherical hardcarbonmaterials from a sugar solution was developed for the firsttime by us.[16] The sugar molecules are first dewatered to formmicelles in solution, and further dewatering leads to theappearance of nuclei-oligomers within the micelles in the formof tiny particles. These nuclei grow gradually into nanome-ter-scaled (size of micelles) spherules by consumption of themicelles. Through polymerization of the grown nuclei withdewatered sugar dissolved in an aqueous phase, spherules ofmicrometer-size are finally formed until all the sugar has beendepleted. The above scheme for the formation of HCS is shown inFigure 1.[17] The following carbonization process has little effecton the morphology of spherules, consequently mono-dispersedHCS are obtained, as shown in Figure 2.

The obtained HCS materials show a perfect sphericalmorphology with a smooth surface. High-resolution transmis-

m 4593

REVIE

W

www.advmat.de

Xuejie Huang is currently aprofessor at the Institute Phy-sics, Chinese Academy ofSciences (IOP-CAS). Hereceived his B.Sc. degree(1986), M.Sc. degree (1989),and Ph.D. degree (1993) fromXiamen University of China,University of Science andTechnology of China, and DelftUniversity of Technology of theNetherlands, respectively. In1994–1995, he worked as a

postdoctoral fellow at the Technische Fakultat of theChristian–Albrecht University in Kiel, Germany. Since 1996,he has been the leader of the Group of Solid State Ionics atIOP-CAS. His research interests are focused on solid stateionics and rechargeable batteries.

4594

sion electron microscopy (HRTEM) and X-ray diffraction (XRD)investigations indicate that the interior part of the particle has adisordered structure. It cannot be graphitized at 2500 8C.[17] HCSis rich in 0.4–0.8 nm micropores. The specific surface area isabout 400m2 g�1. This value can be increased by activation anddecreased after chemical vapor deposition (CVD) treatment. Themicropore size is influenced by the preparation method andcondition. For example, the average pore size is 0.45 nm forHCS1 obtained by a normal hydrothermal method, while HCS2with smaller micropores (0.39 nm) was obtained by a micro-emulsion-mediated hydrothermal method.[18]

The particle size of the HCS can be controlled by adjusting theconcentration. The maximum particle obtained is about 5mmwhen the sugar solution is 5 M. The size was reduced to 250 nmwhen thea concentration of the sugar solution was decreased to0.15 M.[17] HCS with diameters of 100–200 nm (nano-HCS) couldbe also prepared through a modified hydrothermal method withpolyacrylamide (PAM) as additive,[19] as shown in Figure 2b.

HCS have also been prepared by other authors from glucoseand starch.[20,21] They have been used as templates forsynthesis,[22] catalyst supports,[23–25] counter electrodes fordye-sensitized solar cells (DSSCs),[26] and as anodes for Li-ionbatteries,[16–19] as shown below.

Li storage in HCS shows a typical feature of hard carbon asshown in Figure 2c and 2d. The voltage profile is composed of twoparts: one is the slope region at a higher voltage (>0.09V), and theother is the flat region at lower voltage (<0.09V). The first part isvery similar to the voltage profile of coke,[4,13] and is related tolithium in the buckled graphene sheets and edges. The secondpart is ascribed to the filling of lithium in the micropores, whichoccurs at a lower voltage. Stevens and Dahn confirmed thismechanism in pyrolyzed glucose near 0V (vs Li/Liþ) by means ofin-situ small-angle X-ray scattering techniques.[27] It is not

Figure 1. Scheme of the preparation of HCS from a sugar solution by the hydReproduced with permission from [17]. Copyright 2002, Elsevier.

� 2009 WILEY-VCH Verlag Gmb

possible to see the few lithium atoms in the micropores byimaging techniques, however, it might be possible to see lithiumgrains or clusters in the large pores. We performed HRTEMexperiments to observe lithium storage in pores directly for thefirst time. For this purpose, a special carbon nanotube materialwith a bamboo-like large cavity, polymerized carbon nitridenanobells (CNNBs), was selected as a model material.[28,29]

CNNBs were fully lithiated by discharging to 0.0 V vs Li/Liþ. Asshown in Figure 3 b and 3c, it is noticed that some tinynanocrystallites with a stripe spacing in the range of0.25–0.29 nm are found inside the tube. Correspondingly, thereare some scattered diffraction dots in the inset selected area

rothermal method.

H & Co. KGaA, Weinheim

electron diffraction (SAED) pattern com-pared with the initial material. The nano-crystallites are distributed in the cavitiesinside the tube and are rarely found inthe surface regions and the tube walls. It canbe concluded that the nanocrystallites arelocated on the inner surfaces instead ofdeposited on the outer surfaces of the tubewalls. These tiny grains disappear afterdelithiation (Fig. 3d). The fringe spacingsof lithium–graphite intercalation compounds(Li-GICs) cannot drop in the above rangesaccording to the lattice parameters and thedistribution of the nanosized crystallites. Atambient conditions, the stable lattice struc-ture of lithium is body centered cubic (bcc),and the corresponding spacing is 0.248 nmfor the (110) plane. Hence the stripes in thesamples were ascribed to metallic lithiumnanocrystallites.[30,31]

It should be mentioned that outside thetube wall, there is a uniform amorphous filmwith a thickness of about 5–7 nm. That is theso called solid electrolyte interphase (SEI)layer formed by the reduction of the electro-lyte.[32,33] It is also noticed that the thickness

Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

Figure 2. Scanning electron microscopy (SEM) images of a) HCS0 (normal hydrothermalmethod 1000 8C, 2 h) [16], and b) nano-HCS (1000 8C, 2 h) obtained by a reverse micellehydrothermal method [19]. Charge/discharge curves of c) HCS0 (1000 8C, 2 h) and d) nano-HCS(1000 8C, 2 h). Reproduced with permissions from [16] and [19]. Copyright 2001 and 2007,Elsevier.

Figure 3. a) TEM image of CNNBs. b) HRTEM image of fully lithiatedCNNBs (discharged to 0 V), in region (I) as marked in (a). c) HRTEM imageof fully lithiated CNNBs (discharged to 0 V vs Li/Liþ), in region (II) asmarked in (a). d) HRTEM image of fully delithiated CNNBs (charged to3.5 V vs Li/Liþ). Reproduced with permission from [31]. Copyright 2003,ECS.

Figure 4. Cyclic performance of HCS anode materials. a) HCS1 (1000 8C,2 h, micropore size r¼ 0.46 nm), cycled between 0.0mV and 2.0 V.b) HCS2 (1000 8C, 2 h, micropore size r¼ 0.39 nm), cycled between0.0mV and 2.0 V [18]. c) Nano-HCS, cycled between 0.0mV and 2.0 V.d) Carbon-coated nano-HCS, cycled between 0.0mV and 2.0 V.e) Nano-HCS, cycled between �20mV and 2.0 V. f) Carbon-coatednano-HCS, cycled between �20mV and 2.0 V.

of the SEI film at a delithiated state (Fig. 3d) was thinner than thatat the full lithiated state (Fig. 3b). The thickest part is only 3 nmand some places were almost naked. This means that the SEI wasdecomposed at high voltage during the charging process. Thisfinding is against the normal impression that that the SEI on acarbon anode is stable.

Li storage in nano-HCS has also been investigated byHRTEM.[19] Although we did not observe inserted lithium, itwas found that about 4–6 layers of parallel graphene sheets appearafter full lithiation. It seems that disordered nano-HCS becomeslightly ordered. In order to clarify this, Raman spectra of HCS

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Wein

materials with different hydrogen contentwere recorded after different cycle numbers.It was found that the La value of thenano-HCS1 with more hydrogen (H/C¼ 0.14 at./at.) is increased for the first fivecycles while the La value of the nano-HCS2with less hydrogen ((H/C¼ 0.07 at./at.) doesnot vary significantly. A Li-replace-H mechan-ism was proposed to explain this disorder-to-order phenomenon. This result indicatesthat the host atoms, i.e., carbon, may bemobileand rearrange to be ordered to a certain levelupon lithiation and delithiation.

Li storage in micropores occurs at a very lowvoltage around 0V vs Li/Liþ. Differentmaterials and electrodes have different polar-ization situations. Therefore, the Li-storagecapacity in HCS was significantly influencedby the cut-off voltage. As shown in Figure 2d,there is a large irreversible capacity loss fornano-HCS, caused mainly by the formation ofa thick SEI film. The existence of an insulatingSEI filmmay lead to an increase of the internalresistance of the electrode. Consequently, thevoltage plateau for Li storage in microporesdoes not appear for nano-HCS. When the

cut-off voltage was extended to �20mV, or by coating aconductive carbon layer on it, larger Li-storage capacities atlow voltage are achieved, as shown in Figure 4e and 4f.

The kinetic performances of HCS materials were alsoinvestigated.[18] It was found that the capacities at a highervoltage (0.09–2V) almost do not vary with the current density. Butthe capacities at a lower voltage (0–0.09V) are influenced stronglyby the current density. This indicates that Li storage inmicroporesnear 0V shows poor kinetic performance.

heim 4595

REVIE

W

www.advmat.de

Figure 5. HRTEM images of fully lithiated nano-SnO (7.8 Li). a) Lowmagnification and b,c) different regions inside the particle. Reproducedwith permission from [39]. Copyright 1998, ECS.

4596

2.2. Tin-based Oxides

In 1995, Idota et al. of the Fuji Film Co. claimed a class ofamorphous tin-based composite oxides (TCOs) as anode activematerials.[34,35] These materials showed 500–600mA h g�1

reversible capacities. At that time, it was announced that SnII�Ois the active center for lithium insertion and other glass-formingelements, which make up an oxide network. 7Li NMRmeasurements evidenced the high ionic state of lithium retainedin the lithiation state. Idota et al. believed that there was a newLi-storage mechanism. This finding attracted great attention.

According to in-situ XRD results reported by Dahn’s group andRaman spectroscopy studies performed in this group,[36,37] it isclear that lithium storage in tin-based oxides occurs by a two-stepmechanism. Taking SnO as an example, the reaction equationscan be written as:

2Liþ SnO ! Li2Oþ Sn (1)

4:4Liþ Sn $ Li22Sn5 (2)

This mechanism was also supported by the fact that fivereversible redox peaks can be observed in the cyclic voltammo-gram of the SnO anode, which is hard to be explained by othermechanisms.[38]

It was well known that Li�Sn alloy reactions in a stand-aloneSn electrode showed a very poor cyclic performance since theformation of Li22Sn5 leads to a significant volume expansion andthe electrode tends to pulverize during cycling. Therefore,Courtney and Dahn proposed that the Li2Omatrix acts as a glue toretard the aggregation of tin atoms into large coherent regions inthe case of tin oxides.[36] A direct image of a fully lithiatednano-SnO anode was given in 1998 by us for the first time.[39] Asshown in Figure 5, nanometer-SnO maintains its particle shapeafter full lithiation (1551mA h g�1) but the interior part wasconverted into a nanocomposite where Li�Sn crystallites(2–20 nm) are dispersed within an amorphous Li2O matrix. Thisresult shows a clear picture that the enhanced cyclic performanceof the alloy reaction in the case of tin-based oxides could benefitedfrom two structural factors: alloy grains are on the nanometerscale and are dispersed in another inactive phase. The formationof nanocomposites becomes an important strategy for the designof high-capacity anode materials as discussed later.

As shown in Figure 5, a layer of thin film (1.6 nm) can be seenclearly on the surface of the reacted nanoparticles. As analyzed bythe results of FTIR spectroscopy,[39] it is the SEI film, similar tothe SEI film on a carbon anode.

After Fuji’s finding, great efforts have been paid to exploreoxide anodes that contain active elements able to form an alloywith lithium. Since Reaction (1) is normally irreversible, theinitial efficiency is less than 70% in most of cases, researchers’attentions were shifted to alloy materials, such as Si, Sn, and Sb,or alloy-based composites.

Before further discussion on the alloy anode, it is worthmentioning that Reaction (1) is partially reversible, as confirmedrecently by Xue and Fu.[40]

� 2009 WILEY-VCH Verlag Gmb

2.3. Nanosized Alloy Anodes

The electrochemical alloying reaction of lithium with metals hasbeen widely studied since the 1970s.[41–43] Many metals andalloys can store a large quantity of lithium by the formation ofalloys (Li4.4Si, which corresponds to a Li storage capacity of4200mA h g�1, Li4.4Ge: 1600mA h g�1, LiAl and Li4.4Sn:990mA h g�1, and Li3Sb: 665mA h g�1). Themain difficulties forusing alloy-based materials are their dramatic volume expansionand contraction during Li insertion and extraction. This leads tothe pulverization of the electrode materials and poor cyclicperformance.[42,43] Before the report on TCO, it was reported byYang et al. that superfine Sn(SnSb)0.14 (200 nm) showed improvedcyclic performance. It was supposed that small particles have astronger endurance to volume variation.[44]

These results enlightened us that nanosized alloys may havebetter cyclic performances. In 1999, we reported the electro-chemical performance of silicon nanoparticles (SiNPs) and Sinanowires (SiNWs) for the first time.[45,46] Pure nanometer-scaleSi powder (80 nm) was prepared by laser-induced silane gasreaction.[47] Si nanowires were synthesized by laser ablation.[48]

At room temperature, SiNPs and SiNWs show reversiblecapacities of 1700 and 900mA h g�1, respectively (Fig. 6a). It wasfound that the amount of carbon black plays a key role to improvethe capacity retention, as shown in Figure 6b. This means that theelectronic contact between the active particles is very importantfor SiNPS to achieve better cyclic performance. However, thesame strategy is not effective for micrometer-sized Si particles(mm-Si). It was observed that serious pulverization occurs for themm-Si electrode.[49] SiNPs show a significant advantage incapacity retention towards large-size particles based on this result,which is consistent with our expectation.

Previous studies revealed that Li can form various compoundswith Si at elevated temperatures, such as Li12Si7, Li13Si4, Li7Si3,and Li22Si5.

[50,51] However, multiple voltage plateaus cannot beseen in the voltage profiles of SiNPs and SiNWs in Figure 6. Asshown in Figure 7c and 7d, it was found that electrochemicallithiation into Si at room temperature led to the formation of anamorphous product.[46,49] We suggested that the insertion of Li

H & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

Figure 6. a) Discharge/charging curves of a) SiNPs (weight ratio of SiNPs/carbon black¼ 1: 1) [39] and b) SiNWs (no conductive additive) [46]. b) Thecyclability of Si-based anodes at different conditions. The cell is: Si/1 M

LiPF6, EC/DEC (1: 1)/Li. The current density for curve 5 is 0.8mA cm�2,others are 0.1mA cm�2. Curve 1) mm-Si/CB (4: 4), 0.0–0.8 V, 2) mm-Si/CB(4: 4), 0.0–2.0 V, 3) SiNPs/CB (4: 4), 0.0–2.0 V, 4) SiNPs/CB (9: 1), 0.0–0.8V,5) SiNPs/CB (4: 4), 0.0–0.8V, and 6) SiNPs/CB (4: 4), 0.0–0.8 V. Repro-duced with permission from [45]. Copyright 1999, ECS.

Figure 7. a) SEM image of SiNPs. b) HRTEM image of SiNWs. A layer ofSiOx covers the surface of the SiNWs. c) HRTEM image of lithiated SiNPS.d) HRTEM image of lithiated SiNWs. e) SEM image of lithiated SiNPelectrode (with 1: 1 carbon black). f) SEM image of lithiated SiNP electrode(no carbon black). White arrow: Cu substrate. Black arrow: agglomerateddense Si layers. Reproduced with permissions from [45] and [49]. Copyright1999, ECS and 2000, Elsevier, respectively.

into Si leads to a distorted angular bond distribution. With anincrease in doping dose, the bond distortion becomes moreobvious and at last destroys the silicon structure. At roomtemperature, sluggish diffusion of Si inhibits the formation of anordered Li�Si alloy phase. It was observed that a more orderedstructure of lithiated SiNWs appeared after annealing at 400 8Cfor 5 h under vacuum conditions.[49]

It was also found that serious agglomeration occurs after onecycle, as shown in Figure 7e and 7f.[49] Nanosizedmaterials have ahigh surface energy. They tend to form large agglomerates,however, in most of cases, nanoparticles do not merge together atroom temperature because of slow transport kinetics of the hostatoms and poor contact. During electrochemical lithiation, theparticles are expanded, which increases their contact probability.After delithiation, most Si atoms are in a dangling bond state withhigh energy. They tend to bond together if no other atom canbond with them or prevent their transport, as shown in the case ofSnO. Consequently, very dense agglomerates are formed afterelectrochemical cycling (see Fig. 7f). We called this phenomenon‘electrochemical agglomeration’. It has been observed in most ofthe alloy anodes mentioned later.

Because of the difficulty of synthesizing nanosized silicon atthat time, a SnSb alloy attracted our attention. Besenhard et al.prepared a Sn0.78Sb0.22 alloy (0.56Snþ 0.22SnSb) with a particle

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag G

size of about 200 nm by electroplate and reductive precipita-tion.[44,52,53] We developed a reductive co-precipitation method.Dendrite-like nanosized pure rhombohedral phase b-SnSb alloyswere formed around 0 8C.[54]

The cyclability of nano-SnSb was obviously better thanindividual Sn and Sb anodes.[54,55] As proposed by Besenhardet al., reacted SnSb may be embedded in the unreacted soft andductile Sn matrix.[44] The SnSb alloy can be regarded as an active/active composite at the atomic level.

It was interesting to find that several redox peaks appeared inthe cyclic voltammogram of the nanosized SnSb alloy anode. Itseems that Li�Sb alloy reactions occur first during lithiation, andthen followed the Li�Sn alloy reactions. This was confirmedclearly by ex-situ XRD results.[54]

Since the initial nano-SnSb alloy is a single-phasematerial. Thesuccessive phase transitions mean that inserted lithium firstreacts with Sb locally. In this step, Sb atoms are enriched locally toform Li2Sb and Li3Sb domains. During this step, separated Snatoms may rearrange to form Sn domains. After Li3Sb is formed,the inserted lithium will react with Sn domains to form a series ofLi�Sn alloys. Upon delithiation, the reverse steps occur.Separated Sn or Sb atoms or domains tend to rearrange torestore the original b-SnSb alloy structure. This result provides anexample that reverse phase separation and restoration canoccur at room temperature driven by electrochemical reactions.It means that the diffusion of Sn, Sb, and Li at roomtemperature should be very fast to follow the structurevariation and the structure of the b-SnSb alloy phase is athermodynamic favorable phase. The details of the local structure

mbH & Co. KGaA, Weinheim 4597

REVIE

W

www.advmat.de

Figure 9. A) Voltage profile of nano-SnSb/HCS. B) Voltage profile of nano-SnSb/MCMB.C) Cyclic performance of a) nano-SnSb/MCMB, b) nano-SnSb/HCS, and c) nano-SnSb.D) Columbic efficiency of a) nano-SnSb/MCMB, b) nano-SnSb/HCS, and c) nano-SnSb.Reproduced with permission from [56] and [57]. Copyright 2001, RSC and 2002, ACS, respectively.

4598

evolution, in addition to electron distribution,are still not understood clearly.

Similar to the Si anode, serious electro-chemical agglomeration occurs for nano-SnSballoys.[55] In order to solve this problem, ananosized SnSb alloy was deposited on thesurface of MCMB and HCS respectively,[56,57]

as shown in Figure 8.The cyclic performance of the nano-SnSb/

HCS composite is improved significantlycompared with nano-SnSb, as shown inFigure 9. However, it is found that thecolumbic efficiencies for these materials ineach cycle are not very high (<99%), and this isnot acceptable for real Li-ion batteries. Thereason for the low efficiency is related to thedirect exposure of the nanosized alloy particleto the electrolyte. Due to significant volumevariation, the SEI film on the alloy particlesmay not be very stable during cycling, whichleads to continuous formation of the SEI infreshly exposed areas that appear in each cycle.This problem could be solved by a strategy ofembedding nanosized alloy particles in acarbon matrix[58–62] or shell.[63,64] The core–

shell strategy has also been used successfully for Si anodes.[65–68]

Consequently, a high columbic efficiency (initial efficiency>85%, efficiency for each cycle >99%), high reversible capacity(>600mA h g�1), and excellent cyclic performance, as well ashigh tap density (>1 g cm�3) and low surface area (<5m2 g�1)were achieved.

2.4. Transition Metal Oxides

In 2000, Poizot et al. reported for the first time that lithium can bestored reversibly in transition metal (TM) oxides through a

Figure 8. SEM images of a,b) SnSb/MCMB and c,d) SnSb/HCS. Repro-duced with permissions from [56] and [57]. Copyright 2001, RSC and 2002,ACS, respectively.

� 2009 WILEY-VCH Verlag Gmb

heterogeneous conversion reaction:[69]

Liþ TMO ! Li2Oþ TM (3)

where TM is Co, Fe, Ni, and Cu. Later, reversible lithium storagewas also observed in TM fluorides, sulfides, nitrides, andphosphides.[70–76] It is very interesting in view of fundamentalresearch that very inert LiF or Li2O can react with a TM at roomtemperature. This was unexpected previously. Actually, in 1999,reversible lithium storage in CuO and Fe2O3 had been reported byus.[32] However, we could not explain this behavior at that time. Itis clear now that the enhanced electrochemical reactivity of LiF orLi2O is mainly a benefit of the special microstructure where theconverted LiX and TM components show an extremely smallgrain size (<5 nm) and intersperse with each other uni-formly.[69,70,77,78] The very short diffusion lengths and largecontact areas in nanocomposites are kinetically favorable forunusual reversible electrochemical behaviors of LiX/TM nano-composites.

In view of increasing the energy density of the batteries,besides a high capacity, a low average charging (delithiation)voltage of the anodematerial is another important factor. A typicalvalue of the average charging voltage at a low rate is 0.3 V forgraphite, 0.5 V for silicon, 0.7 V for Sn, and 1.6 V for Li4Ti5O12. Inaddition, a small polarization (voltage difference betweencharging and discharging curve) of less than 0.2 V is necessaryto achieve a high energy efficiency. This requirement is realizedfor anode and cathode materials by the intercalation or alloymechanism. However, as for TM compounds that undergo aconversion reaction, most of the materials suffer from a highoverpotential (voltage difference between the working voltage andthermodynamic equilibrium voltage) for both lithium insertionand extraction (about 1V).[79] Consequently, a theoretical

H & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

Figure 10. HRTEM of nano-Cr2O3 particles in the: a,b) fully lithiated stateand c,d) fully delithiated state.

lithiation capacity could be approached only when the thermo-dynamic equilibrium voltage (also called electromotive force,emf) of the material for the conversion reaction is higher than1V.[79] Cr2O3 and MnO are two transition metal oxides that have ahigh lithiation capacity but relatively low emf values (1.085 and1.032V vs Li/Liþ, respectively).[79] They are more suitable asanode materials for Li-ion batteries.

Rhombohedral Cr2O3 (85-0869) has a theoretical density of5.235 g cm�3 and shows a dark green color. Commercial Cr2O3

powder material shows a particle size of 200–500 nm and theelectronic conductivity is about 1.78� 10�7 S cm�1.[80] It has beenknown that materials with a high electronic conductivity canachieve a high columbic efficiency for conversion reactions, asshown in the case of RuO2.

[78] Mg and Ni doping can improve theconductivity of Cr2O3.

[81–84] Therefore, Mg and Ni-doped Cr2O3

were also prepared. Their electronic conductivities wereimproved to 3.11� 10�4 and 9.57� 10�5 S cm�1, respectively,after 2mol % doping.[80]

The initial discharge capacities of Cr2O3 materials are1200–1400mA h g�1 and are mainly contributed from a plateauregion at 0.15V.[80,85] This capacity is larger than the theoreticalvalue of 1058mA h g�1 for Cr2O3 (6Li needed for a full reductionof Cr2O3 into a 3Li2O/2Cr composite). The initial chargecapacities are 700–800mA h g�1 in the voltage range from 0to 3.0 V and the voltage profile shows mainly two sloped regions.As discussed in recent papers for such behavior,[70,78,79,86] thesloped region at a low voltage region could be assigned extra Listorage at interfacial regions of the LiX/TM nanocomposite. Inthe case of the Cr2O3 system, it contributes a Li-extraction capacityof about 250mA h g�1. The sloped region above the emf value forthis reaction is related to the phase transformation from the Li2O/Cr to Cr2O3, as confirmed by the XPS results.[85] The chargingcapacities are about 500mAh g�1, which aremuch lower than thetheoretical capacity for a complete Li-extraction reaction. Itindicates that almost half of the inserted lithium is trappedirreversibly.

The average charging voltage for Cr2O3 is about 1.2 V, which ismuch lower than most of the other TM compounds (TMX, inbrief) (>1.8 V). This is a significant advantage of Cr2O3 as ananode material compared with other transition metal oxides andfluorides, because of its low emf value of 1.058V vs Li/Liþ.

After full insertion, all particles are covered by a layer ofamorphous species with a thickness of 20–90 nm (Fig. 10a).[80]

This layer has been observed in many lithiated TMX anodes andrepresents the so-called SEI. In a zoomed image in Figure 10b,obvious disintegration of Cr2O3 grains is observed. Some tinygrains of less than 5 nm are dispersed in amorphous regions.Such a microstructure is similar to most other TM oxide anodesobserved.[77,78] After full delithiation, the interior part convertsback into a more ordered Cr2O3 phase (Fig. 10d) and the SEIthickness is decreased to less than 5 nm, some particles are evennaked. Similar to previous observations,[69,77,78] the SEI on theTM oxide anodes are electrochemically decomposable. Sincethe thick SEI covers all particles, it is believed that the reversibleformation/decomposition of the SEI film is not favorable forachieving good cyclic performance. It may lead to the peeling offof active materials from the current collector.[85]

The SEI film is normally regarded as a passivating layer, whichdoes not decompose during charging and is around several

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag G

nanometers thick in the cases of graphite carbon anodes.[87] Theformation of the SEI will be terminated when the electron cannottunnel through this film. Why the SEI film is so thick on TMXanodes and what the possible components are, are two importantissues. Recently, Gachot et al. studied a chromium-based oxide(named CBO) grown on stainless steel mesh tissues as an anodein a half cell with 1M LiPF6 dissolved in ethylene carbonate (EC)/dimethyl carbonate (DMC) (1/1, w/w) as electrolyte. A layer of a20–50 nm amorphous film was observed on a fully lithiated CBOelectrode. After analyzing the supernatant in a borosilicatemicro-fiber (Millipore) film separator recovered from Li/CBOcells, the existence of MeOLi, CH3OCO2Li, and differentpoly(ethylene oxide) (PEO) oligomer series were identified clearlybased on a high-resolution mass spectrometry (ESI-HRMS)comparison analysis of a series of reference compounds.[88] Inorder to determine the SEI components on the electrode, inparticular the electrochemically decomposable components, atechnique that combined thermogravimetry and mass spectro-metry (TG-MS) was applied[89] that was first used by Zhao et al. toanalyze the SEI film on a lithiated graphite electrode.[90] It wasfound that the Cr2O3 electrode in the fully lithiated state shows acontinuous weight loss of up to 8.97% at 500 8C. While theelectrodes charged at 1.1, 2.0, and 3.0 V show a 5.36%, 1.7%, and1.51% weight loss at 500 8C, respectively. Since the maininorganic species in the SEI film, i.e., Li2CO3, LiF, and lithiatedproducts of Cr, Li2O, and unreacted Cr2O3, cannot be decom-posed below 500 8C, the thermally decomposable components inthe electrode are mainly organic or polymer species in the SEIfilm. Accordingly, the electrode in the fully lithiated state shouldhave the largest quantity of organic and/or polymer-like species.These results mean that the thermal decomposable species in thefully lithiated Cr2O3 electrode are decomposed electrochemicallyupon charging. The weight ratio of the SEI film in the Cr2O3

electrode is decreased by at least 7.5wt. % after charging to 3.0 V.

mbH & Co. KGaA, Weinheim 4599

REVIE

W

www.advmat.de

Figure 11. Cyclic performance of Cr2O3-based anodes. a) Pristine Cr2O3,b) Cr2O3 with 10wt. % binder, c) nano-sized Cr2O3, d) 2wt. % Mg-dopedCr2O3, e) carbon-coated Cr2O3, f) carbon-coated Cr2O3/carbon blackmixture followed by a further CVD treatment, g) CNTs grown on aCr2O3 nanocomposite, and h) carbon-coated Mg-doped Cr2O3.

Figure 12. Polarization analysis (DV curve) of anodematerials at a low rate(0.05–0.2 C). a) Nano-HCS, b) SnO, c) Si, d) nano-SnSb, e) Cr2O3, f) MnO,g) MCMB, and h) Li4Ti5O12.

4600

This consists with the significant thinning of the SEI film on theCr2O3 anode after charging as observed by TEM.

Based on mass spectrometry analysis, it was found thatoligomers and polymers with ethylene oxide units are electro-chemically decomposable species during charging, and ROCO2Liis electrochemically relatively stable. The formation of a largeamount of these species on lithiated Cr2O3 could be related to astrong catalytic effect from the nanosized TM nanograins, whichneeds further clarification. It was also found that the TG curves ofMCMB and HCS are similar as the delithiated Cr2O3 samples.This means that much less thermal decomposable species areformed on the carbon electrode, which is consistent with Zhaoet al.’s TG result that the SEI films on graphite is thin. In addition,it was found that ROCO2Li could be the main thermaldecomposable component in the SEI layers on MCMB andHCS electrodes.[89]

The cyclic performance of a pristine Cr2O3 electrode was verypoor.[85]We supposed that themain reasons were a result of a loss ofelectronic contact of active particles and poor transport propertiesfor electrons and ions within the active material. The first issuecould be a result of a large volume variation of the electrode layer,which could be solved by decreasing the particle size, coating, andavoiding the formation of an unstable SEI film. The solution forsolving the second issue could also be decreasing particle size, inaddition to improving the intrinsic electronic conductivity byhetero-atom doping and keeping good contact with a conductiveadditive. Several material modifications that aim to solve the aboveproblems have been performed,[80,85] including decreasing theparticle size, hetero-atom doping, and preparation of a carbon/Cr2O3 composite with different microstructures.

For samples (a) to (d) in Figure 11 the surface of the Cr2O3

particles is exposed to the electrolyte phase. Therefore, such amicrostructure cannot avoid the formation of the unstable SEIfilm on the Cr2O3 particles. Carbon coating is very effective toimprove the cyclic performance, as shown in the case of samples(e) and (h) in Figure 11. Carbon nanotubes (CNTs) grown on theCr2O3 particle to form a composite where each Cr2O3 particle isfully covered by coiled CNTs, is also effective to improve the cyclicperformance. However, such an urchin structure cannot avoid theformation of a SEI on the exposed area of Cr2O3. The sample (f)shows the best cyclic performance. Cr2O3 particles were firstmixed with carbon black, and then the composite particles werecoated with a layer of carbon. After that, the particles were treatedby CVD to cover all the exposed area. Such a composite has theadvantage of maintaining electronic contact and avoiding theformation of an unstable SEI film during cycling, as evidencedpartially by TEM investigation.[80]

For anodematerials based on the conversion reaction, there arethree main disadvantages, which are low columbic efficiency,high average charging voltage, and high polarization.

The polarizations of different anodes have been compared bythe DV curves. The DV curve is obtained by subtracting thedischarging curve of the second cycle from the charging curve ofthe first cycle after normalization. It is the sum of theoverpotentials of both charging and discharging at a galvanostaticmode. The DV curve method was used to roughly compare thepolarization of different materials.[91,92] As shown in Figure 12,the average polarization order is Li4Ti5O12<MCMB< Si<nano-SnSb< SnO<HCS<MnO<Cr2O3. Obviously, lithium

� 2009 WILEY-VCH Verlag Gmb

storage in the conversion reaction shows the highest polarization.The main origin seems related to either Xn� or Mnþ transportproperties, and is not significantly dependent on the transportof electrons and lithium ions. Among TM oxides, MnOshows a low polarization, high capacity (450–550mA h g�1,3500–4200mA h cm�3), and good rate performance.[92] Recently,MgH2 was tested as an anode material, and underwent a similarconversion reaction mechanism. It was found that the averagecharging voltage was about 0.5 V and the DV is about 0.5 V at0.05C.[93] This finding seems to support the viewpoint thattransport of anions or cations from the host is a key factor todetermine the polarization. How to decrease the high polarizationof the conversion reaction is still a great challenge for most of theTMX materials.

H & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 4593–4607

RE

www.advmat.de

3. Cathode Materials

VIE

W

Figure 13. a) TEM image of Al2O3-coated nano-LiCoO2 heat-treated at 300 8C for 2 h. The fringespacing is 0.472 nm, close to the d-value of 0.468 nm of the (003) plane [JCPDS 77-1370]. b) Plotsof discharge capacities of A) bare commercial LiCoO2, B) Al2O3 coated on LiCoO2, heated at300 8C for 2 h, and C) Al2O3 coated on LiCoO2, heated at 600 8C for 2 h as a function of the cyclenumber for the first 20 cycles. The cut-off voltage is from 3.0 to 4.5 V. Reproduced withpermission from [105]. Copyright 2002, Elsevier.

3.1. Surface Coating on LiCoO2 and

LiMn2O4

With decades of study, the research on thecathode materials for lithium-ion batteries hasbeen focused on layered, structured hexagonalLiCoO2, spinel LiMn2O4, olivine LiFePO4, andtheir derivatives. LiCoO2 is the first cathodematerial that has been used in commercialLi-ion batteries.[4,94,95]

LiCoO2 cathode materials are typicallycycled between the fully lithiated dischargestate LiCoO2 (�3.0 V vs Li/Liþ) and a roughlyhalf-delithiated charge state LixCoO2

(x¼ 0.5–0.6, 4.2 V vs Li/Liþ) to yield a useablespecific capacity below 150mA h g�1. When

the Li/LiCoO2 cell is cycled within a limited composition range of0< x< 0.5, it shows excellent cycling performance. However, itscapacity fades rapidly when more Li is deintercalated from thelattice. Reimers and Dahn believed that overcharging leads tocathode degradation and electrolyte decomposition at highvoltage.[96] As lithium is removed from LixCoO2, Co

3þ is oxidizedto unstable Co4þ. A high concentration of Co4þ will damage thecathode crystallinity and result in a significant decrease of thec-axis dimension caused by a phase transition. The contractionalong the c-axis results in mechanical failure in LiCoO2 particlesand rapid capacity fading.

Besides the interior lattice structural stability, it was evidencedthat the performance degradation of LiCoO2 is related to thedissolution of its Co3þ ions in the electrolyte solution.[97] Aurbachet al. reported that the electrochemical behaviors of LixMOy

(M¼Ni, Mn) cathode materials were strongly dependent on theirsurface chemistry.[98]

In order to improve structure stability, one early strategy was tocoat the LiCoO2 surface with inert oxides such as SnO2, Al2O3,and MgO, by Kweon and Park and Cho et al. in 2000.[99–101]

Among these, Al2O3-coated LiCoO2 exhibits excellentcapacity retention. It was suggested that the formation of asolid solution of LiCo1 – yAlyO2 improved the structural stability ofLiCoO2 during cycling.[100] However, many studies have beencarried out to substitute Co with Al or Mg to form a solid solutionand the capacity fading was still significant during cycling and thestructural stability was not improved obviously.[102,103]

We developed a simple co-precipitation method to coat Al2O3

or MgO on commercial LiCoO2 (Nippon Chemical Industry) andnano-LiCoO2 in 2002.[104–106] As shown in Figure 13a, a 4 nmamorphous Al2O3 layer was coated on nano-LiCoO2 uniformly.According to XRD patterns, the lattice of LiCoO2 did not change.It means that the Al or Mg did not dope into the LiCoO2 lattice inthe obtained sample.

We also demonstrated, as did Kweon and Park and Cho et al.,that both Al2O3 and MgO surface coating improves the cyclingstability of commercial LiCoO2 significantly when the chargevoltage extends to 4.5 V vs Liþ/Li. Figure 13b shows the effect ofan Al2O3 coating on improving the cyclic performance.[105]

The understanding of the coating mechanism was argued atthat time. Cho et al. suggested that the effectiveness of the surface

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag G

coating was attributed to the suppression of the variation of thec-value of LiCoO2 in the case of the ZrO2 coating, which leads to azero-strain material.[107] Chen and Dahn argued that the ZrO2

coating does not affect the variation of lattice parameters duringcycling through their in-situ XRD studies.[108]

In order to clarify this, in-situ XRD data was collected on bothAl2O3 coated and uncoated LiCoO2 multiply cycled in the voltagerange between 3 and 4.8 V. The 4.8 V high-voltage limit waschosen to accelerate the structural damage of the uncoatedsample.

We performed in-situ XRD studies at the National SynchrotronLight Source (NSLS) at the Brookhaven National Laboratory, incollaboration with Yang et al.[109] For brevity, Figure 14 showspartial results of the in-situ XRD patterns during discharge. It wasfound that the XRD patterns of commercial LiCoO2 remainunchanged during the whole discharge process (Fig. 9 inRef. [109]), which indicated that the bulk structure of thecommercial LiCoO2 cathodematerial did not reversibly transformback from H2 (the phase at the end of a 5.2 V over-charge) to H1(the phase of commercial LiCoO2). When the Al2O3-coatedLiCoO2 cathode was charged to 5.2 V vs Liþ/Li, the in-situ XRDpatterns indicate that the material experiences a series of phasetransitions, from H1 to H2, and then to O1a, and finally to O1during the initial charge process. In the subsequent dischargeprocess from 5.2 to 3 V, the phase transitions are fully reversible(Fig. 14b).

In addition to other results, the in-situ XRD analysis indicatesclearly that the variation range of the lattice parameters is muchlarger in the Al2O3-coated LiCoO2 than those in the pristinesamples. Considering that the capacity fading of the Al2O3-coatedsample is much smaller than that for the pristine cathode, webelieve that surface coating does not suppress the phasetransition during over-charging.

These results explain that the coating is effective in view ofmaintaining the structural stability. However, it was not under-stood why the coating could maintain the structural stability.Many authors believe that the surface coating improves thecycling stability of LiCoO2 by physically separating the oxidizedactive material from the electrolyte.[110–113]

A comparison of the XPS results of the uncoated andMgO-coated LiCoO2 at fully charged states seems to support the

mbH & Co. KGaA, Weinheim 4601

REVIE

W

www.advmat.de

Figure 14. In-situ XRD patterns of a pristine (left, a) and Al2O3-coated (right, b) LiCoO2 cathodeduring the first discharge from 4.7 to 2.0 V at a D/2.9 rate in the (003) to (102) region (for thepristine) and from 5.2 to 3.0 V at a D/3.9 rate in the (003) to (105) region. Reproduced withpermission from [109]. Copyright 2004, ECS.

4602

above suggestions.[114] Figure 15 shows O 1s spectra of the coatedand uncoated LiCoO2 charged at various voltages. The electronicstructure of oxygen for uncoated LiCoO2 varies steadily (at529.4 eV in uncharged commercial LiCoO2 electrode) with theincrease of the charge voltage and a new component appears at532.6 eV, which corresponds to oxygen atoms with a strongeroxidizing power. The content of such oxygen atoms increases withthe charge voltage and becomes dominant in the electrode at highvoltages. Oxygen atoms with a higher binding energy (at 531.6 eV)in the coated cathode, however, appear at a lower charge voltagebut their content increases more slowly than in the uncoatedcommercial LiCoO2 cathode (Fig. 15b). Moreover, as the bindingenergy of these oxygen atoms in the charged MgO-coated LiCoO2

cathode is lower than in charged commercial LiCoO2, theiroxidizing power should also be weaker than those in the

Figure 15. Comparison of O 1s spectra of commercial (a) and MgO-coated (b) LiCoO2 chargedto various voltages. Reproduced with permission from [114]. Copyright 2003, ECS.

commercial LiCoO2 cathode. These resultsindicate that modifying the surface of theLiCoO2 particles may be helpful to suppressthe release of oxygen, which may lead toinstability of the interior structure and decom-position of the electrolyte.

It was noticed in 2003 by us that LiCoO2 canreact with the electrolyte spontaneously.[115]

Based on GC-MS analysis, the soakage ofLiCoO2 in EC/DMC solvents leads to therelease of carbon dioxide (CO2), carbonmonoxide (CO), methane (CH4), ethylene(C2H4), water (H2O), ethane (C2H6), andoxygen (O2), in the order of their maximumcounts per second from high to low. FTIRspectroscopy detected the existence of ROLiand ROCO2Li, and XRD detected the existenceof Co2O3 and Co3O4. Later, we observeddirectly the existence of the surface film onnanosized LiCoO2,

[116] as shown in Figure 16.

� 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinhe

As analyzed from corresponding SAEDpatterns,[116] the Co3O4 phase appears aftersoaking in the solvent and in the electrolyte.ROCO2Li was detected from the FTIR results.These results confirm that the surface ofLiCoO2 can react with the organic solventspontaneously, especially for nanosizedLiCoO2. Actually, we did find that nanosizedLiCoO2 showed the worst cyclic performanceat 10C, as shown in Figure 17. Whennanosized LiCoO2 was sintered at 850 8Cand grown to a micrometer-sized particle, itshowed a much improved cyclic performanceat 10C. As also shown in Figure 17, when theaged commercial LiCoO2 was annealed at850 8C, it also showed good cyclic performanceat 10C. These results indicate that surfacespecies exist on the uncoated samples (agedcommercial product and nanosized LiCoO2),which are not good for achieving good cyclicand rate performance. Coating or annealingcan improve their properties by either forminga more stable surface or by decreasing theactive surface area.

Another important issue is whether coated inert oxides arestable during storage and electrochemical cycling in Li-ionbatteries. Kim et al. found that the ZrO2 coating layer is convertedinto ZrF4 in non-aqueous electrolyte.[110] This is because of theexistence of HF in the LiPF6-based electrolyte. It was found by usthat directly adding nanosized Al2O3 to the electrolyte was alsovery effective to improve the cyclic performance of aged LiCoO2.Li3AlF6 and AlF3 was found on LiCoO2 after soaking in thenano-Al2O3-added electrolyte.[117] These results indicate that theoxides for coating, such as Al2O3, MgO, and ZrO2

, may not bestable in a LiPF6-based electrolyte that contains a trace amount ofH2O. Theymay convert into fluorides finally to stabilize the activesurface of LiCoO2. The formation of surface fluorides could beconverted from coated oxides, or formed through complicatedchemical reactions when Al3þ or other cations exist in the

im Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

Figure 16. TEM images of nanosized LiCoO2 particles at different situ-ations. a) Pristine, b) soaked with 1M LiPF6 dissolved in EC/DMC (1: 1, v/v)electrolyte for 7 days, c) soaked with anhydrousDMC (<10 ppm) for 7 days,and d) after 10 cycles in a LiCoO2/Li cell over a voltage range of 2.5–4.3 V.Reproduced with permission from [116]. Copyright 2006, ECS.

Figure 17. Cyclic performance of LiCoO2 at 2.5–4.3 V at 10C. a) Nano-LiCoO2, b) nano-LiCoO2, 850 8C, 6 h, c) nano-LiCoO2, 650 8C, 24 h,d) commercial LiCoO2, e) nano-LiCoO2, 850 8C, 24 h, and f) commercialLiCoO2, 850 8C, 24 h. Electrolyte: 1 M LiPF6, EC/DMC (1: 1).

LiPF6-based electrolyte. The possibility to form a surface layerdriven by a chemical reaction may explain the facts that even aloose and incomplete oxide coating can still improve theperformance of the cathode materials,[118–120] even when thecoverage was as low as 13.7%.[120]

Recently, Sun et al. reported that a uniform 10nm coating ofAlF3 on LiCoO2 can improve the capacity retention at 3–4.5 Vsignificantly.[121] They have extended this strategy to stabilizeother layer-structured cathodes.[122,123] These achievementsindicate that the coating effect could be more effective whenthe coating phase is stable in a non-aqueous electrolyte. Thisneeds further investigation. One question still remains: when thecoating layer is dense and uniformly covered on the surface of

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag G

each particle of cathode material, how could electrons transportthrough a 10 nm insulating coating layer? Further careful studiesshould be performed to clarify this issue.

Enlightened by the achievement of the coating on improvingthe stability of LiCoO2, we applied a similar strategy toLiMn2O4.

[124,125] LiAlO2 was coated on LiMn2O4 and formed asurface solid solution. The coated material showed an excellentcyclic performance, rate performance, and excellent capacityretention at 55 8C.[125]

3.2. Research on LiFePO4

Iron-based compounds that contain compact tetrahedral ‘poly-anion’ structural units (XO4)n (X¼ S, P, As, W, or Mo) have beeninvestigated intensively as potential cathode materials for lithiumion batteries.[8,126–130] The elements in the compounds areabundant in the earth, inexpensive, and environmentally friendly.In particular, orthorhombic LiFePO4 with an ordered olivinestructure has attracted the most attention. However, this kind ofcompound is a wide-gap semiconductor (3.7 eV) and has aninherently extremely low electronic conductivity (�10�9 S cm�1)at room temperature. Various material processing approacheshave been adopted to overcome this drawback, includingmethodsto coat the phosphate particles with carbon.[131,132] Chung et al.found that controlled cation non-stoichiometry combined withsolid solution doping bymetals supervalent to Li, e.g., Mg, Zr, andNb, increases the electronic conductivity of LiFePO4 by a factor of108.[133] However, the origin of the enhanced conductivity isunder debate. Controversy is focused on whether the supervalentions are actually doped into the lattice and the improvedconductivity may be caused by a conductive nanonetwork.[134,135]

Here, our efforts on doping are reviewed.The first-principles calculation, which has already made an

impact on the understanding of practical lithium-ion batterymaterials after Cedar et al.’s work,[136–141] was used by us for pureLiFePO4 and its delithiated counterpart FePO4. Experimentally, adoping system was studied by Goni et al.[142] who found that Fe3þ

can substitute some of the Liþ ions in the LiMgPO4 structure toform the solid solution Li1 – 3xFexMgPO4(0, x, 0.1), to createcation vacancy channels along a certain crystallographic direction.Based on this experimental illumination, the electronic structureof Cr-doped LiFePO4, namely, Li1 – 3xCrxFePO4 with x¼ 1/32, hasbeen calculated from first-principles by a generalized gradientapproximation (GGA) in order to elucidate the underlyingconducting mechanism.[143] However, the obtained result did notgive a proper bandgap.[143] In order to correct this, the calculationwas performed again using the Vienna ab initio simulationprogram within the DFTþU framework proposed by Zhouet al.,[144] and core electrons were treated with the ProjectorAugmented Wave method. A U value of 4.3 eV was used for Feand the bandgap is well reproduced as shown in Figure 18.

As for doping, GGAþU calculations show that although Crdoping introduces impurity states, the bandgap is still as large as2 eV. Therefore, the possibility of p-type conduction by impuritystates can be eliminated. Electron transport is probably in theform of small polaron hopping.[145] As the nearest Li�Fe distancein LiFePO4 is 3.28 A, while the nearest Fe�Fe distance is 3.87 A,electron hopping between Fe and Cr in Li sites might be easier. It

mbH & Co. KGaA, Weinheim 4603

REVIE

W

www.advmat.de

Figure 18. Density of states of LiFePO4, Li5/8Cr1/8FePO4, LiFe7/8Na1/8PO4,and LiFePO4 – 1/4N1/4, calculated from GGAþU. The Fermi level is set as areference.

4604

is found that for Li5/8Cr1/8FePO4, electronic states at the Fermilevel mainly come from the Cr 3d, but contain hybridization withneighboring O 2p and Fe 3d bands of the O and Fe atoms near thedopant. This Cr-induced state may alter the conductivity withrespect to the pure LiFePO4, where no electronic state is located atthe Fermi level.

Following the theoretical calculation, we also prepared theCr-doped LiFePO4 samples.[143] It was found that its conductivitywas enhanced after substituting a small amount of lithium ions.In Figure 19a, it can be seen that doped Li0.97Cr0.01FePO4 andLi0.91Cr0.03FePO4 show an electronic conductivity much higherthan that of pure LiFePO4 at room temperature. It should benoted that the activation energy Ea¼ 0.186 eV of pure LiFePO4

obtained by us in a previous study is even smaller than the one(0.5 eV) given by Chung et al.[133] for Mg, Zr, and Nb-dopedsamples. It was recognized later that the sample could be dopedwith Mg because one of the raw materials contained Mgimpurities, the content of which was around 1% as determinedlater by inductively coupled plasma (ICP) analysis.

Pure phase LiFePO4 was prepared again recently. High purityraw materials were used to eliminate the impurities, i.e., Mg, asdetermined by ICP. There was no carbon detected by Raman

igure 19. a) Direct current (DC) electrical conductivity of Li0.97Cr0.01FePO4 and Li0.91Cr0.03FePO4

ompared with LiFePO4 (with �1% Mg impurities). Reproduced with permission from [143].opyright 2003, APS. b) DC electrical conductivity of Li0.97Fe0.96Na0.05PO4 – d and LiFePO4 (usingigh purity precursors).

FcCh

� 2009 WILEY-VCH Verlag Gmb

spectroscopic analysis. The dc-conductivity of pure phaseLiFePO4 was measured again using the same method asdescribed in ref. [143]. It shows a conductivity of nearly 10�9 Scm�1 at room temperature with an activation energy of 0.639 eV.

The diffusion mechanism of Li ions in the olivine LiFePO4 wasinvestigated by us from first-principles calculations.[146] Theenergy barriers for possible spatial hopping pathways werecalculated with the adiabatic trajectory method. The calculationsshow that the energy barrier along the c-axis was about 0.6 eV.However, the other migration pathways have much higher energybarriers that result in a very low probability of Li-ion migration.This means that the diffusion in LiFePO4 is 1D. The 1D diffusionbehavior has also been confirmed by a full ab initio moleculardynamics simulation, through which the diffusion behavior wasdirectly observed.[146]

Because of the feature of 1D ionic transport, it was necessary toknow whether the Cr ions in the Li sites could block the diffusionof Li ions along the 1D diffusion pathway.[147] We performed abinitio density functional theory (DFT)-based calculations usingthe Vienna ab initio simulation package VASP.[148–151] This codesolves the Kohn–Sham equations within the pseudopotentialapproximation whereby the electrons are described in thelocal-density approximation (LDA) by ultrasoft pseudopoten-tials.[152,153] The valence electrons are expanded in a plane wavebasis set and the effect of the core states on the valence electronsis treated with ultrasoft pseudopotentials. The structural energyminimization was performed for pure LiFePO4 and Li29/32Cr1/32FePO4 in a 2� 2� 2 super-cell. The Monkhorst–Pack schemewith 7� 3� 5 k point sets was used for the integration in theirreducible Brillouin zone. The energy cut-off for the plane waveswas chosen to be 600 eV.[154] In the structure optimization, partialoccupancy at the Fermi level was treated according to Methfesseland Paxton.[155] In the case of Li29/32Cr1/32FePO4, there are two Livacant sites in the super-cell; the optimized configuration was thatthe two vacancies lie nearby the Cr ions along the c-axis direction.The optimized configuration was obtained through the following:place the Cr ion in one Li site and change the locations of the twovacancies to construct new configurations. For each configurationthe total energy was calculated after relaxation has occurred. Theconfiguration with lowest total energy was considered to be theoptimized one.

In order to determine the migration energy barriers of the Liand Cr ions in the LiFePO4 crystal, we employed the so-called

H & Co. KGaA, Weinheim

‘adiabatic trajectory method’ after structuraloptimization.[156,157] The energy barrier wasobtained conveniently through monitoringthe changes of the total energy. The migrationenergy barrier was only considered as themigration of a Li ion from one occupied site toa nearby vacant site. When we considered themigration of one Li ion, we had already‘created’ one vacant site in front of the movingLi ion. This was the same for the migration ofthe Cr ions with the vacancies pair (threevacancies in all). Figure 20a presents themigration energy barriers obtained by mon-itoring the total energy changes during themovement of the migration ions. It can beseen from the figure that themigration energy

Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

Figure 20. a) Migration energy barriers of Li and Cr ions in the LiFePO4 crystal along the 1Ddiffusion pathway. b) Migration energy barriers of Li ions in the LiFePO4 and LiFePO4 – 1/4N1/4

along the 1D diffusion pathway. Reproduced with permission from [147]. Copyright 2004, IOP.

barriers along the c-direction for Li ions and Cr ions are about 0.6and 2.1 eV, respectively. The different migration energy barriersfor Li ions and Cr ions indicate clearly that the Li ions can diffusealong the c-direction while the Cr ions are not diffusible and canonly be kept oscillating at their initial positions. It is, therefore,understandable that the enhanced electronic conductivity as aresult of the substitution of a small amount of high valence metalions for lithium ions will not lead to improved electrochemicalperformance. The reason lies in the blocking of the 1D pathwaysby the heavy high-valence metal ions in the Li sites. The positivecontributions due to the enhanced electronic conductivity arepartially offset by the negative blocking effect of the Cr ions onlithium ion motion.

Since doping at the Li site is not favorable for the diffusion ofLi, doping at the Fe or O site was considered. We proposed thenew strategies of Na-doping at Fe sites and N-doping at Osites.[158–160] First-principle calculations have been performed toidentify the doping effect on the electron structure. Calculationswere carried out within the DFTþU framework, and coreelectrons were treated with the projector augmented wavemethod. Two unit cells were used for calculation. The density ofstates (DOS) of LiFe7/8Na1/8PO4 is shown in Figure 18. Theimpurity states appear when Na is doped, which leads to a muchnarrower bandgap of 0.7 eV. An enhanced electron transportproperty is expected in Na-doped LiFePO4 according tocalculations. As shown in Figure 19b, the Na-doped materialLi0.97Fe0.96Na0.05PO4 – d shows a much improved electronicconductivity and the activation energy Ea¼ 0.035 eV is also muchsmaller. Small polaron hopping mechanism is accepted inLiFePO4.

[145,161] Electron transport is dependent upon smallpolaron hopping of Fe3þ holes within the lattice. Na-dopingintroduces corresponding Fe3þ, one of the occupied Fe3d statesgets unoccupied after doping and this is the origin of the impuritystates.

Replacement of O by N could also introduce Fe3þ. The DOS ofN-doped LiFePO4 is also shown in Figure 18.[162] New statesappear in the bandgap of pure LiFePO4 after N doping, andthe bandgap drops from 3.7 to 1.3 eV. The new states areinduced by the Fe�N bonding. Fe3þ formed as a result ofelectron redistribution. The hybridization of Fe 3d and N 2p leadsto a narrower bandgap. When N is doped, it induces Fe3þ sitesand electron-hole hopping along the Fe�N�Fe chains mayoccur.

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Wein

As N locates at the nearest neighbor site ofthe Li migration pathway and it has a differentvalence state compared with O, N doping mayinfluence the Li activation energy. The ‘elasticband’ method was used to calculate the Liactivation energy. The activation energy for Lidiffusion in LiFePO3.75N0.25 is slightly lowerthan that in LiFePO4.

[162] This is caused by thedistortion of the Fe�O(N) octahedral, but itdoes not significantly affect the activationenergy for Li diffusion. The above calculationsdemonstrate the feasibility of Na-doping at Lisites or N-doping at O sites in LiFePO4.

4. Summary and Outlook

Investigations on spherical hard carbon and polymerized carbonnitride nanobells demonstrate that a large quantity of lithium canbe stored in carbon nanopores. When lithium was inserted,nanometer-SnO was converted into a nanocomposite whereLi�Sn crystallites (2–20 nm) are dispersed within an amorphousLi2O matrix. Nanometer-scale Si showed a very high capacity andimproved cycling performance. Surface coating improves thecycling stability of LiCoO2 when its charge limit voltage isextended to 4.5 V vs Liþ/Li, which does not suppress the phasetransition during over-charging, rather, it suppresses the releaseof oxygen. Na-doping into Fe sites and the partial replacement ofoxygen by nitrogen in LiFePO4 were proposed, which is differentfrom other approaches and was demonstrated theoretically as aneffective strategy to achieve high electronic conductivity but notblock the transport of lithium ion along the 1D channels.

Based on a decade’s research, the performances of Li-ionbatteries are being improved by introducing new and enhancedchemical combinations. Graphite anode could be replaced withmetals and oxides such as Si, Sn, Sb, SnO, Cr2O3, or MnObecause of their high capacity. Severe crystallographic volumechanges of high capacity materials during the lithiation anddelithiation process could be solved by using nanosized ornanostructured materials. However, serious side reactionsobserved inmost stand-alone nanomaterials need to be controlledfor real battery applications. Combining the advantages ofnanometer and micrometer materials, surface modification,the introduction of new additives for the electrolyte, and theintroduction of polymer-based electrolytes, are crucial to achieve astable electrode/electrolyte interface, which is very important forthe safety and service life of a battery. Therefore, the integrationstrategy, considering the electrode materials, electrode structure,and the electrode/electrolyte interface, is essential for futureresearch. In particular, for LiFePO4, transport of electrons andions in LiFePO4 has been well investigated by theoreticalcalculations. However, experimental work based on singlecrystals and thin films is needed to obtain precise data forfurther confirmation and clarification. New synthetic methods ateto be developed to introduce different dopants into LiFePO4. Thegreat interest in phosphates may not only lead to highperformance cathode materials for Li-ion batteries but alsomay shed some light on new applications as electronicmaterials.

heim 4605

REVIE

W

www.advmat.de

4606

Acknowledgements

Financial support from NSFC (50672122, 50730005, 60621061), ‘863’project (2006AA03Z346, 2006AA03Z228) and ‘973’ project(2007CB936501) are appreciated. Contributions from Dr. Q. Wang,Dr. L. H. Shi, Dr. X. D. Wu, Dr. L. J. Liu, Dr. J. Hu, Dr. D. Y. Wang, Dr. Y. C.Sun, Dr. S. Q. Shi, Dr. C. Y. Ouyang, Dr. X. J. Wang, Dr. N. Liu, Dr. J. Y. Liu,and Mr. Z. J. Liu to the experimental part are appreciated.

Received: May 22, 2009

Published online: September 14, 2009

[1] J. Tollefson, Nature 2008, 456, 436.

[2] Y. Takayuki, K. Hisayuki, A. Masanori EP 391281, 1990.

[3] T. Nagaura, K. Tozawa, Prog. Batt. Solar Cells 1990, 9, 209.

[4] Y. Nishi, Chem. Record 2001, 1, 406.

[5] A. S. Arico, P. Bruce, B. Scrosati, J. Marie, Tarascon, W. V. Schalkwijk,Nat.

Mater. 2005, 4, 366.

[6] J. Maier, Nat. Mater. 2005, 4, 805.

[7] M. M. Thackeray, P. J. Johnson, L. A. De Picciotto, P. G. Bruce,

J. B. Goodenough, Mater. Res. Bull. 1984, 19, 179.

[8] A. K. Padhi, K. S. Nanjundaswamy, J. B. Goodenough, J. Electrochem. Soc.

1997, 144, 1188.

[9] M. Armand, Materials For Advanced Batteries, (Eds: D. W. Murphy,

J. Broodhead, B. C. H. Steele), Plenum Press, New York 1980.

[10] K. Xu, Chem. Rev. 2004, 104, 4303.

[11] J. Yamaura, Y. Ozaki, A. Morita, A. J. Ohta, J. Power Sources 1993, 43, 233.

[12] A. Mabuchi, Tanso 1994, 165, 298.

[13] J. R. Dahn, T. Zheng, Y. H. Liu, J. S. Xue, Science 1995, 270, 590.

[14] H. Azuma, H. Imoto, S. Yamada, K. Sekai, J. Power Sources 1999, 81, 1.

[15] J. L. Tirado, Mater. Sci. Eng. R 2003, 40, 103.

[16] Q. Wang, H. Li, L. Q. Chen, X. J. Huang, Carbon 2001, 39, 2211.

[17] Q. Wang, H. Li, L. Q. Chen, X. J. Huang, Solid State Ionics 2002, 152, 43.

[18] J. Hu, H. Li, X. J. Huang, Solid State Ionics 2005, 176, 1151.

[19] J. Hu, H. Li, X. J. Huang, Solid State Ionics 2007, 178, 265.

[20] X. M. Sun, Y. D. Li, Angew. Chem. Int. Ed. 2004, 43, 597.

[21] X. J. Cui, M. Antonietti, S. H. Yu, Small 2006, 2, 756.

[22] Y. S. Hu, Adelhelm, Philipp, Smarsly, M. Bernd, H. Sarmimala, A. Markus,

J. Maier, Adv. Funct. Mater. 2007, 17, 1873.

[23] R. Z. Yang, X. P. Qiu, H. R. Zhang, J. Q. Li, W. T. Zhu, Z. X. Wang,

X. J. Huang, L. Q. Chen, Carbon 2005, 43, 11.

[24] R. Z. Yang, H. Li, X. P. Qiu, L. Q. Chen, Chem. Eur. J. 2006, 12, 4083.

[25] Z. H. Wen, J. Liu, J. H. Li, Adv. Mater. 2008, 20, 743.

[26] Z. Huang, X. Z. Liu, K. X. Li, D. M. Li, Y. H. Luo, H. Li, W. B. Song,

L. Q. Chen, Q. B. Meng, Electrochem. Commun. 2007, 9, 596.

[27] D. A. Stevens, J. R. Dahn, J. Electrochem. Soc. 2000, 147, 4428.

[28] X. D. Bai, D. Y. Zhong, G. Y. Zhang, X. Ma, S. Liu, E. G. Wang, Y. Chen,

D. T. Shaw, Appl. Phys. Lett. 2001, 79, 1552.

[29] X. Ma, E. G. Wang, W. Zhou, D. A. Jefferson, J. Chen, S. Deng, N. Xu,

J. Yuan, Appl. Phys. Lett. 1999, 75, 3105.

[30] D. Y. Zhong, G. Y. Zhang, S. Liu, E. G. Wang, Q. Wang, H. Li, X. J. Huang,

Appl. Phys. Lett. 2001, 79, 3500.

[31] Q. Wang, H. Li, L. Q. Chen, X. J. Huang, D. Y. Zhong, E. G. Wang,

J. Electrochem. Soc. 2003, 150, A1281.

[32] E. Peled, J. Electrochem. Soc. 1979, 126, 2047.

[33] D. Aurbach, J. Power Sources 2000, 89, 206.

[34] Y. Idota, M. Nishima, Y. Miyaki, T. Kubota, T. Miyasaki, EU 0651450a1,

1995.

[35] Y. Idota, T. Kubota, A. Matsufuji, Y. Maekawa, T. Miyasaka, Science 1997,

276, 1395.

[36] I. A. Courtney, J. R. Dahn, J. Electrochem. Soc. 1997, 144, 2045.

[37] W. F. Liu, X. J. Huang, Z. X. Wang, H. Li, L. Q. Chen, J. Electrochem. Soc.

1998, 145, 59.

[38] H. Li, X. J. Huang, L. Q. Chen, Solid State Ionics 1999, 123, 189.

� 2009 WILEY-VCH Verlag Gmb

[39] H. Li, X. J. Huang, L. Q. Chen, Electrochem. Solid-State Lett. 1998, 1,

241.

[40] M. Z. Xue, Z. W. Fu, Electrochem. Solid-State Lett. 2006, 9, A468.

[41] A. N. Dey, J. Electrochem. Soc. 1971, 118, 1547.

[42] R. A. Huggins, J. Power Sources 1999, 81, 13.

[43] M. Winter, J. O. Besenhard, M. Spahr, P. Novak, Adv. Mater. 1998, 10, 725.

[44] J. Yang, M. Winter, J. O. Besenhard, Solid State Ionics 1996, 90, 281.

[45] H. Li, X. J. Huang, L. Q. Chen, Z. G. Wu, Y. Liang, Electrochem. Solid-State

Lett. 1999, 2, 547.

[46] G. W. Zhou, H. Li, H. P. Sun, D. P. Yu, Y. Q. Wang, X. J. Huang, L. Q. Chen,

Z. Zhang, Appl. Phys. Lett. 1999, 75, 2447.

[47] Y. L. Li, Y. Liang, K. S. Xiao, F. Zheng, Z. Q. Hu, Acta Metall. Sin. 1995, 31,

B21.

[48] D. P. Yu, C. S. Lee, I. Bello, X. S. Sun, Y. H. Tang, G. W. Zhou, Z. G. Bai,

Z. Zhang, S. Q. Feng, Solid State Commun. 1998, 105, 403.

[49] H. Li, X. J. Huang, L. Q. Chen, G. W. Zhou, Z. Zhang, D. P. Yu, Y. J. Mo,

Solid State Ionics 2000, 135, 181.

[50] C. J. Wen, R. A. Huggins, J. Solid-State Chem. 1976, 37, 271.

[51] K. Amezawa, N. Yamamoto, Y. Tomii, Y. Ito, J. Electrochem. Soc. 1998,

145, 1986.

[52] J. O. Besenhard, J. Yang, M. Winter, J. Power Sources 1997, 68, 87.

[53] J. Yang, M. Wachtler, M. Winters, J. O. Besenhard, Electrochem. Solid-State

Lett. 1999, 2, 161.

[54] H. Li, G. Y. Zhu, X. J. Huang, L. Q. Chen, J. Mater. Chem. 2000, 10, 693.

[55] H. Li, L. H. Shi, W. Lu, X. J. Huang, L. Q. Chen, J. Electrochem. Soc. 2001,

148, A915.

[56] L. H. Shi, H. Li, Z. X. Wang, X. J. Huang, L. Q. Chen, J. Mater. Chem. 2001,

11, 1502.

[57] H. Li, Q. Wang, L. H. Shi, L. Q. Chen, X. J. Huang, Chem. Mater. 2002, 14,

103.

[58] R. L. Turner, D. J. Mcclure, L. J. Krause, M. M. Buckett, J. R. Dahn, O. Mao,

US 6 203 944 2001.

[59] S. Kawakami, M. Asao, US 6 949 312, 2005.

[60] S. Mizutani, H. Inoue, US Pat. Appl 0208378, 2005.

[61] S. Mizutani, Y. Kudo, US Pat. Appl 0121348, 2006.

[62] A. D. W. Todd, R. E. Mar, J. R. Dahn, J. Electrochem. Soc. 2007, 154, A597.

[63] K. T. Lee, Y. S. Jung, S. M. Oh, J. Am. Chem. Soc. 2003, 125, 5652.

[64] W. M. Zhang, J. S. Hu, Y. G. Guo, S. F. Zheng, L. S. Zhong, W. G. Song,

L. J. Wan, Adv. Mater. 2008, 20, 1160.

[65] T. Umeno, K. Fukuda, T. Tsunawaki, Y. Hara, JP 2000346090, 2000.

[66] M. Yoshio, H. Y. Wang, K. Fukuda, T. Umeno, N. Dimov, Z. Ogumi,

J. Electrochem. Soc. 2002, 149, A1598.

[67] H. Li, J. Hu, X. J. Huang, L. Q. Chen, CN ZL 1328805c, 2004.

[68] H. Li, X. J. Huang, CN ZL 100422112c, 2005.

[69] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon,Nature 2000,

407, 496.

[70] H. Li, G. Richter, J. Maier, Adv. Mater. 2003, 15, 736.

[71] F. Badway, N. Pereira, F. Cosandey, G. G. Amatucci, J. Electrochem. Soc.

2003, 150, A1209.

[72] F. Badway, F. Cosandey, N. Pereira, G. G. Amatucci, J. Electrochem. Soc.

2003, 150, A1318.

[73] P. Poizot, S. Laruelle, S. Grugeon, L. Dupont, J. M. Tarascon, J. Electro-

chem. Soc. 2002, 149, A1212.

[74] N. Pereira, L. C. Klein, G. G. Amatucci, J. Electrochem. Soc. 2002, 149,

A262.

[75] D. C. C. Silva, O. Crosnier, G. Ouvrard, J. Greedan, A. Safa-Sefat,

L. F. Nazar, Electrochem. Solid-State Lett. 2003, 6, A162.

[76] Y. Wang, Z. W. Fu, X. L. Yue, Q. Z. Qin, J. Electrochem. Soc. 2004, 151,

E162.

[77] A. Debart, L. Dupont, P. Poizot, J. B. Leriche, J. M. Tarascon, J. Electro-

chem. Soc. 2001, 148, A1266.

[78] P. Balaya, H. Li, L. Kienle, J. Maier, Adv. Funct. Mater. 2003, 13,

621.

[79] H. Li, P. Balaya, J. Maier, J. Electrochem. Soc. 2004, 151, A1878.

H & Co. KGaA, Weinheim Adv. Mater. 2009, 21, 4593–4607

REVIE

W

www.advmat.de

[80] J. Hu, H. Li, X. J. Huang, L. Q. Chen, Solid State Ionics 2006, 177,

2791.

[81] A. Holt, P. Kofstad, Solid State Ionics 1997, 100, 201.

[82] A. Holt, P. Kofstad, Solid State Ionics 1999, 117, 21.

[83] I. Ayub, F. J. Berry, C. Johnson, D. A. Johnson, E. A. Moore, X. Ren,

H. M. Widatallah, Solid State Commun. 2002, 123, 141.

[84] D. Conran, D. Lonergan, J. Phys. Chem. Solids 1977, 38, 333.

[85] J. Hu, H. Li, X. J. Huang, Electrochem. Solid-State Lett. 2005, 8, A66.

[86] J. Jamnik, J. Maier, Phys. Chem. Chem. Phys. 2003, 5, 5215.

[87] R. Fong, U. V. Sacken, J. R. Dahn, J. Electrochem. Soc. 1990, 137,

2009.

[88] G. Gachot, S. Grugeon, M. Armand, S. Pilard, P. Guenot, J.-M. Tarascon,

S. Laruelle, J. Power Sources 2008, 178, 409.

[89] Y. Zeng, L. F. Li, H. Li, X. J. Huang, L. Q. Chen, Ionics 2009, 15, 91.

[90] L. W. Zhao, I. Watanabe, T. Doi, S. Okada, J. Yamaki, J. Power Sources 2006,

161, 1275.

[91] J. P. Sun, K. Tang, X. Q. Yu, J. Hu, H. Li, X. J. Huang, Solid State Ionics 2008,

179, 2390.

[92] X. Q. Yu, Y. He, J. P. Sun, K. Tang, H. Li, L. Q. Chen, X. J. Huang,

Electrochem. Commun. 2009, 11, 791.

[93] Y. Oumellal, A. Rougier, G. A. Nazri, J. M. Tarascon, L. Aymard, Nat.

Mater. 2008, 7, 916.

[94] M. S. Whittingham, Chem. Rev. 2004, 104, 4271.

[95] T. Ohzuku, J. Power Sources 2007, 174, 449.

[96] J. N. Reimers, J. R. Dahn, J. Electrochem. Soc. 1992, 139, 2091.

[97] G. G. Amatucci, J. M. Tarascon, L. C. Klein, Solid State Ionics 1996, 83, 167.

[98] D. Aurbach, K. Gamosky, B. Markovsky, G. Salitra, Y. Gofer, U. Heider,

R. Oesten, M. Schmidt, J. Electrochem Soc. 2000, 147, 1322.

[99] H. J. Kweon, D. G. Park, Electrochem. Solid-State Lett. 2000, 3, 128.

[100] J. Cho, C. S. Kim, S. I. Yoo, Electrochem. Solid-State Lett. 2000, 3, 362.

[101] J. Cho, Y. J. Kim, B. Park, Chem. Mater. 2000, 12, 3788.

[102] H. Tukamoto, A. R. West, J. Electrochem. Soc. 1997, 144, 3164.

[103] Y. I. Jang, B. Huang, H. Wang, D. R. Sadoway, G. Ceder, Y. M. Chiang,

H. Liu, H. Tamura, J. Electrochem. Soc. 1999, 146, 862.

[104] Z. X. Wang, C. Wu, L. J. Liu, F. Wu, L. Q. Chen, X. J. Huang, J. Electrochem.

Soc. 2002, 149, A466.

[105] L. J. Liu, Z. X. Wang, H. Li, L. Q. Chen, X. J. Huang, Solid State Ionics 2002,

152, 341.

[106] Z. X. Wang, L. J. Liu, L. Q. Chen, X. J. Huang, Solid State Ionics 2002, 148,

335.

[107] J. Cho, Y. J. Kim, T. J. Kim, B. Park, Angew. Chem. Int. Ed. 2001, 40, 3367.

[108] Z. Chen, J. R. Dahn, Electrochem. Solid-State Lett. 2002, 5, A213.

[109] L. J. Liu, L. Q. Chen, X. J. Huang, X. Q. Yang, W. S. Yoon, H. S. Lee,

J. McBreen, J. Electrochem. Soc 2004, 151, A1344.

[110] J. S. Kim, C. S. Johnson, J. T. Vaughey, S. A. Hackney, K. A. Walz,

W. A. Zeltner, M. A. Anderson, M. M. Thackeray, J. Electrochem. Soc.

2004, 151, A1755.

[111] G. G. Amatucci, A. Blyr, C. Sigala, P. Alfonse, J. M. Tarascon, Solid State

Ionics 1997, 104, 13.

[112] E. Endo, T. Yasuda, A. Kita, K. Yamaura, K. Sekai, J. Electrochem. Soc. 2000,

147, 1291.

[113] J. Cho, Y. J. Kim, T. J. Kim, B. Park, J. Electrochem. Soc. 2002, 149,

A127.

[114] Z. X. Wang, X. J. Huang, L. Q. Chen, J. Electrochem. Soc 2003, 150,

A199–A208.

[115] Z. X. Wang, X. J. Huang, L. Q. Chen, J. Electrochem. Soc 2004, 151, A1641.

[116] N. Liu, H. Li, Z. X. Wang, X. J. Huang, L. Q. Chen, Electrochem. Solid-State

Lett. 2006, 9, A328.

[117] J. Y. Liu, N. Liu, D. T. Liu, Y. Bai, L. H. Shi, Z. X. Wang, L. Q. Chen,

V. Hennige, A. Schuch, J. Electrochem. Soc. 2007, 154, A55.

[118] A. M. Kannan, L. Rabenberg, A. Manthiram, Electrochem. Solid-State Lett.

2003, 6, A16.

[119] S. Oh, J. K. Lee, D. Byun, W. I. Cho, B. W. Cho, J. Power Sources 2004, 132,

249.

Adv. Mater. 2009, 21, 4593–4607 � 2009 WILEY-VCH Verlag G

[120] N. Van Landschoot, E. M. Kelder, P. J. Kooyman, C. Kwakernaak,

J. Schoonman, J. Power Sources 2004, 138, 262.

[121] Y. K. Sun, J. M. Han, S. T. Myung, S. W. Lee, K. Amine, Electrochem.

Commun. 2006, 8, 821.

[122] Y. K. Sun, S. T. Myung, B. C. Park, H. Yashiro, J. Electrochem. Soc. 2008,

155, A705.

[123] Y. K. Sun, S. W. Cho, S. W. Lee, C. S. Yoon, K. Amine, J. Electrochem. Soc.

2007, 154, A168.

[124] Z. X. Wang, L. J. Liu, L. Q. Chen, X. J. Huang, CN ZL 01134448.2, 2001.

[125] Y. C. Sun, Z. X. Wang, L. Q. Chen, X. J. Huang, J. Electrochem. Soc. 2003,

150, A1294.

[126] A. Manthiram, J. B. Goodenough, J. Solid State Chem. 1987, 71, 349.

[127] A. Manthiram, J. B. Goodenough, J. Power Sources 1989, 26, 403.

[128] K. S. Nanjundaswamy, A. K. Padhi, J. B. Goodenough, S. Okada,

H. Ohtsuka, H. Arai, J. Yamaki, J. Solid State Ionics 1996, 92, 1.

[129] A. K. Padhi, K. S. Nanjundaswamy, C. Masquelier, S. Okada,

J. B. Goodenough, J. Electrochem. Soc. 1997, 144, 1609.

[130] A. K. Padhi, K. S. Nanjundaswamy, C. Masquelier, J. B. Goodenough,

J. Electrochem. Soc. 1997, 144, 2581.

[131] N. Ravet, Y. Chouinard, J. F. Magnan, S. Besner, M. Gauthier, M. Armand,

J. Power Sources 2001, 97, 503.

[132] N. Ravet, M. Armand, US Pat. Appl 0195591A1, 2002.

[133] S. Y. Chung, J. T. Bloking, Y. M. Chiang, Nat. Mater. 2002, 1, 123.

[134] N. Ravet, A. Abouimrane, M. Armand, Nat. Mater. 2003, 2, 702.

[135] P. S. Herle, B. Ellis, N. Coombs, L. F. Nazar, Nat. Mater. 2004, 3, 147.

[136] G. Ceder, Y. M. Chiang, D. R. Sadoway, M. K. Aydinol, Y. I. Jang, B. Huang,

Nature 1998, 392, 694.

[137] C. Wolverton, A. Zunger, Phys. Rev. Lett. 1998, 81, 606.

[138] V. Meunier, J. Kephart, C. Roland, J. Bernholc, Phys. Rev. Lett. 2002, 88, 075

506.

[139] M. V. Koudriachova, N. M. Harrison, S. W. de Leeuw, Phys. Rev. Lett. 2001,

86, 1275.

[140] J. J. Zhao, A. Buldurn, J. Han, J. P. Lu, Phys. Rev. Lett. 2000, 85, 1706.

[141] S. Q. Shi, D. S. Wang, L. Q. Chen, X. J. Huang, Phys. Rev. B 2003, 67, 115

130.

[142] A. Goni, L. Lezama, A. Pujana, M. I. Arriortua, T. Rojo, Int. J. Inorg. Mater.

2001, 3, 937.

[143] S. Q. Shi, L. J. Liu, C. Y. Ouyang, D. S. Wang, L. Q. Chen, X. J. Huang, Phys.

Rev. B 2003, 68, 195 108.

[144] F. Zhou, C. A. Marianetti, M. Cococcioni, D. Morgan, G. Ceder, Phys. Rev.

B 2004, 69, 201 101.

[145] H. Tukamoto, A. R. West, J. Electrochem. Soc. 1997, 144, 3164.

[146] C. Y. Ouyang, S. Q. Shi, Z. X. Wang, D. S. Wang, L. Q. Chen, X. J. Huang,

Phys. Rev. B 2004, 69, 104 303.

[147] C. Y. Ouyang, S. Q. Shi, Z. X. Wang, H. Li, L. Q. Chen, X. J. Huang, J. Phys. :

Condens. Matter 2004, 16, 2265.

[148] G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.

[149] G. Kresse, J. Hafner, Phys. Rev. B 1994, 49, 14251.

[150] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 1996, 6, 15.

[151] G. Kresse, J. Furthmuller, Phys. Rev. B 1996, 54, 11 169.

[152] M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias, J. D. Joannopoulos, Rev.

Mod. Phys. 1992, 64, 1045.

[153] D. Vanderbilt, Phys. Rev. B 1990, 41, 7892.

[154] H. J. Monkhorst, J. D. Pack, Phys. Rev. B 1976, 13, 5188.

[155] P. E. Blochl, O. Jepsen, O. K. Andersen, Phys. Rev. B 1994, 49, 16 223.

[156] C. Wang, Q. M. Zhang, J. Bernholc, Phys. Rev. Lett. 1992, 69, 3789.

[157] Q. M. Zhang, C. Roland, P. Boguslawski, J. Bernholc, Phys. Rev. Lett. 1995,

75, 101.

[158] D. Y. Wang, L. Q. Chen, H. Li, X. J. Huang, CN CN200410031151.x, 2004.

[159] H. Li, X. J. Huang, D. Y. Wang, L. Q. Chen, CN ZL 200410037502.8, 2004.

[160] H. Li, X. J. Huang, CN ZL200410101618.3, 2004.

[161] T. Maxisch, F. Zhou, G. Ceder, Phys. Rev. B 2006, 73, 104 301.

[162] Z. J. Liu, X. J. Huang, D. S. Wang, Solid State Commun. 2008, 147,

505.

mbH & Co. KGaA, Weinheim 4607